multi-scale tilt depth estimation
TRANSCRIPT
-
7/26/2019 Multi-Scale Tilt Depth Estimation
1/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
2/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
3/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
4/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
5/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
6/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
7/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
8/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
9/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
10/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
11/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
12/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
13/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
14/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
15/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
16/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
17/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
18/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
19/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
20/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
21/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
22/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
23/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
24/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
25/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
26/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
27/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
28/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
29/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
30/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
31/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
32/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
33/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
34/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
35/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
36/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
37/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
38/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
39/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
40/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
41/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
42/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
43/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
44/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
45/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
46/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
47/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
48/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
49/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
50/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
51/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
52/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
53/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
54/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
55/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
56/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
57/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
58/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
59/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
60/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
61/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
62/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
63/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
64/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
65/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
66/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
67/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
68/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
69/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
70/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
71/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
72/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
73/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
74/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
75/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
76/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
77/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
78/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
79/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
80/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
81/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
82/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
83/216
Page 83
achieved by iteratively applying a three element kernel [0.25 0.5 0.25] to
the input data. However, the data is smoothed by an increasing number of
iterations as the continuation distance increases. The number of iterations
increases stepwise and linearly from 1 iteration at the first level of
continuation, up to the number chosen by the user at the final level. Within
the code presented here, both the frequency domain and spatial
smoothing loop may be bypassed by choosing zero iterations.
Figure 16. Example of the frequency domain filter used to low-pass dataprior to tilt computation.
Figure 17. Original noise polluted data and low-pass filtered data at a cut-off wavelength of 100 distance units.
Cosine roll-off
Attenuated coefficients
-
7/26/2019 Multi-Scale Tilt Depth Estimation
84/216
Page 84
Interference
The effect of a regional, non-constant field can in many instances be
equated with interference of the anomalous responses from proximal
sources. The effect of proximal source fields is poorly handled by the
method. The interference at lower frequencies from a fixed multi-source
distribution is more significant than at the higher frequency components.
Due to the methods dependence upon the lower frequency information in
the field, which is retained during upward-continuation, proximal source
fields have a substantial effect on the estimated source depth. This effect
is visibly manifested in the continued distribution, which is noticeablydistorted as per Figure 18.
Figure 18. Linear trend added to synthetic anomalous data produces atrend in the MSTD distribution .
-
7/26/2019 Multi-Scale Tilt Depth Estimation
85/216
Page 85
CHAPTER 3: USE OF THE PROGRAM
The code for the execution of the program, which may be found in the
Appendix A, was compiled in Matlab. The code accepts a 2-column ascii,
comma delimited data file, the name of which must be fully qualified
including the file extension. The format of the columns is such that the 1-
dimensional position of a data point along the profile is specified in the first
column with the corresponding magnetic field intensity in the second
column. The units of distance output by the code will correspond with
those specified in the input data. The assumptions are made that the data
are specified monotonically with respect to distance and that the magnetic
field intensity is in nanotesla. The code makes provision for data specified
at non-regular intervals, interpolating the data onto regularly spaced
positions equal in number and overall profile length to the original data.
Prior to this interpolation, the data are conditioned to remove outliers by
ensuring that within a 3 element kernel, the difference between the centre
point and the median of the kernel is less than the standard deviation of
the same kernel. Care is taken at the extremes of the profile where the twoend points at each end are smoothed by extending the gradients of the
previous two data points as shown in Figure 19.
Figure 19. An example of the conditioning applied at the two endpoints ofthe profile (Solid line connects the conditioned points, dashed lineconnects the original data).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
86/216
Page 86
In order to perform the required frequency domain operations, Fast Fourier
Transforms, herein referred to by the standard abbreviation, FFT, are
applied to the data. Due to the use of FFT processes, the data requires
further conditioning. This conditioning should ensure that the data are
continuous, smooth and differentiable at every point. In order to achieve
this, the data are padded out by one and one half lengths of the original
profile. The data values are inverted about the nearest endpoint value and
in respect of distance. These inverted profiles are attached to each end of
the profile along with additional zeros, half the length of the original profile.
This padded profile is then tapered with the piecewise function, making
use of cosine tapers, shown in Figure 20.
Figure 20. An example of data which has been padded and tapered (Top),the attenuation factor applied to the padded data (Bottom).
Once the data are properly conditioned for the subsequent use of Fourier
type filters, the option is presented to the user to remove a regional field.
The regional field is created via upward continuation of the data by a
distance equal to one eighth of the profile length. This distance was
decided upon empirically based on the assumption that such a
continuation distance would sufficiently separate signals of interest from
Original profile
Tapered padding Tapered padding
-
7/26/2019 Multi-Scale Tilt Depth Estimation
87/216
Page 87
those inadequately represented by the profile given its length. This
regional removal may be done in a more tailored fashion externally and
prior to executing the program. This regional field is then subtracted from
the original data resulting in the residual data utilised from here on. At this
stage the option has been inserted to allow the addition of random white
noise. The amplitude of this noise is determined as a percentage of the
standard deviation of the data. Once the noise component has been
added to the data the entire padded profile is re-tapered using a similar
function to the one shown in the figure above to ensure proper
conditioning at the edges. It should be noted that the frequency of the
noise which is added is controlled at the high end by the data spacing.
The data are now ready for input into the chosen algorithm. Here the
choice is presented to the user to compute the tilt field from the original
total magnetic intensity or from the analytic signal amplitude of the vertical
derivative of the total magnetic intensity (AS(VD(TMI))). This choice, as
elaborated upon within Chapter 2, is due to the method not solving for the
depth to step-like features from the total magnetic intensity. The use of thissecond option is cautioned due to the high order of derivatives employed.
Some care is taken here to reduce the noise amplification by allowing the
user to specify the amount of smoothing prior to the derivative
computations. During the iterative derivations the profiles are also re-
padded and tapered which has been found to be necessary.
The subsequent operation sets up the 2D array of stacked profiles whichwill be transformed into the 2D tilt field. In this step, depth-variable low-
pass filtering is applied both in the frequency domain and spatial domain
as described in Chapter 2. The tilt field is then computed from this filtered
array as per equation 2.1. Now that the data are adequately prepared the
final transformation into the Multi-Scale Tilt Depth (MSTD) field can be
conducted. Each level of the conditioned tilt array is upward continued by
an incremental amount as per equation 2.2. In this code, each level
-
7/26/2019 Multi-Scale Tilt Depth Estimation
88/216
Page 88
represents an increment of 1 distance unit; therefore the number of levels
computed will be equal to the total depth requested by the user and the
problem would have an order proportional to ( total depth). This
topic is highlighted by van Buren (2009), wherein it states that an
improvement in computational efficiency may be had by increasing the
interval of continuation in a non-linear fashion at subsequent levels. This
recommendation was made primarily due to the reduction in vertical
resolution at increased depths. A scheme might be easily implemented
where the continuation interval were increased to 110% of the previous
interval. A simple modification such as this would achieve a four times
increase in speed to compute to a continuation level of 100 distance units
and would have a vertical resolution of 10 units at the last level. Figure 21
illustrates the improvement achieved by such modification.
Figure 21. Non-linear increase in the continuation distance per level toachieve improvements in the speed of computation.
Now that the MSTD distribution is computed the solution locations may be
determined. van Buren (2009) assessed the solution positions by eitherquerying the distribution at the known lateral position of the synthetic
sources, or by solving at the peaks of the analytic signal amplitude of the
input data. This second solution strategy was again attempted but found to
be quite sensitive to noise in the input data, therefore, a new solution
strategy was adopted where the turning point of the zero contour of the
MSTD distribution is sought. This is done by computing the horizontal
derivative of the zero contour and choosing locations where the product of
-
7/26/2019 Multi-Scale Tilt Depth Estimation
89/216
Page 89
adjacent derivative points is negative or zero. A plot of such solution
locations is shown in Figure 22.
Figure 22. MSTD distribution with solution locations determined at theturning point of the zero contour (Above), Zoomed into solution locations(Below).
Late in the development of the solution strategy it was decided to iterate
the entire procedure facilitating the addition of a number of permutations.The user may choose to have a different random noise (with the same
statistical distribution) generated at each iteration, and the dataset run
through the program. A further choice is given to the user, independent of
the noise addition, to increment the smoothing per iteration by a fixed
number of smoothing passes. The combination of these options results in
at least four distinct permutations. The result of each iteration is
assimilated and after all the iterations, the results are plotted as a solution
density image, an example of which may be seen in Figure 23 and Figure
24. The assimilated solutions are also plotted as a histogram of their
depths as in Figure 25. This approach allows the user to gain further
understanding as to the variability of the solution location set with respect
to noise as well as the extent of smoothing. This noise may be chosen to
approximate the amplitude of the noise in the source data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
90/216
Page 90
Figure 23. Solution density image illustrating a high solution density nearthe source s location .
Figure 24. 3D view of solution density illustrating a high solution densitynear the source s location .
Figure 25. A histogram of solution depths illustrating an increased solutiondensity near the source s depth (50 distance units) .
-
7/26/2019 Multi-Scale Tilt Depth Estimation
91/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
92/216
Page 92
Results and Comparison of the Application of Multi-Scale
Tilt Depth Estimation to Synthetic Data
The results of the application of the MSTDE method to the synthetic model
data described in the previous section are collated and presented here.
Summaries of the results of the depth estimation are presented in two
standard graph types (See Figure 31) . The first is a plot of the estimated
depth while the second represents the error in the depth as a percentage
of the true depth of the model. For each of the model types an example of
the model response and various derived products as per Figure 27, as well
as the MSTD distribution and solution densities as per Figure 28, are
shown. Schematics of the various model types are illustrated in Figure 26.
Synthetic model types
Figure 26. Schematics of the various synthetic model types employed inthis study are shown. Single vertical sheet (Top left), Multiple verticalsheets (Top right), Dipping sheet (Bottom left) and Contact in the form of astep (Bottom right).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
93/216
Page 93
Schematic of figures
The following figures represent a standard format for the figures to follow
and are here described.
Figure 27. Schematic of data and various derivatives.
The subplots within Figure 27 contain:
1. A line plot of the input data profile to the program
2. A line plot of the modelled data quantity
3. Horizontal (Solid) and vertical (Dashed) derivatives of the modelled
data quantity
4. The analytic signal amplitude of the modelled data quantity
5. Various upward continuations of the tilt of the modelled data
quantity (distance equal to 0, 25, 50 and 100% of the total sectiondepth)
1
2
3
4
5
-
7/26/2019 Multi-Scale Tilt Depth Estimation
94/216
Page 94
Figure 28. Schematic of data and various section images.
The subplots within Figure 28 contain:
1. A line plot of the input data profile to the program
2. The MSTD distribution plotted as a solid contour plot; The zerocontour is plotted thicker
3. The MSTDE solution density plotted as a colour image with an
arbitrary contour surrounding regions of high solution density for
correlation across subplots
4. The Euler solution density plotted as a colour image with an
instance of MSTDE solutions plotted as circles. The contour from
the MSTDE solution density plot is replicated here
It must be stressed that only the intensities of the fields and not their
geometries are affected by homogeneously varying the unit of distance. In
this work, the generic distance unit was chosen to be the SI distance unit
of 1 meter, resulting in the magnetic fields being reported in nanotesla.
On the use of colour images in order to portray the MSTD distribution and
solution density, the author would like to provide explanation for the
absence of colour bars for each of the images. The intent of the use of the
1
2
3
4
-
7/26/2019 Multi-Scale Tilt Depth Estimation
95/216
Page 95
colour solution density images is to provide more clarity to the diagrams,
especially when presenting solution locations as symbols, the plots would
rapidly degrade into thick lines of solution symbols, preventing the
determination of the densest region of solutions. A clear comparison of
locations from different methods on the same plot would consequently be
hindered. The use of colour density images then facilitates the
determination of the location of the densest region of solutions by showing
the relative solution densities. It is stressed that the location and not the
absolute density of solutions is in fact what is intended to be emphasised
to the reader.
The colour table shown in Figure 29 has been applied to the MSTD
distribution and solution density plots and is illustrated here in order that
the reader may be aware of the relative ordering of colours.
Figure 29. Colour table applied to the MSTD distribution and solutiondensity plots, cool colours on the left are associated with low values andwarm colours on the right are associated with high values of the applicablequantity.
Low High
-
7/26/2019 Multi-Scale Tilt Depth Estimation
96/216
Page 96
Vertical sheet at various depths
Figure 30. Synthetic model data (Depth of 30 dist. units) with plots of the
spatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
97/216
Page 97
MSTDEEuler
Figure 31. Solution depths from MSTDE and Euler (Left), Percentage errorin estimated depths (Right).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
98/216
Page 98
Sheet with various dips
Figure 32. Synthetic model data (60 dip, Depth of 50 dist. units) with plots
of the spatial derivatives, analytic signal amplitude and various continuedtilt products (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
99/216
Page 99
MSTDEMSTDE at model location
Euler MSTDE AS(VD(TMI))
Figure 33. Solution depths from MSTDE, MSTDE at known model location,MSTDE applied to the AS(VD(TMI)) and Euler (Left), Percentage error inestimated depths (Right).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
100/216
Page 100
Vertical sheet with various noise amplitudes
Figure 34. Noise contaminated synthetic model data (Noise range of 10% of
the standard deviation, Depth of 50 dist. units) with no smoothing applied.Plots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
101/216
Page 101
Figure 35. Noise contaminated synthetic model data (Noise range of 10% ofthe standard deviation, Depth of 50 dist. units) with frequency domain andup to 100 iterations of spatial domain smoothing applied. Plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
102/216
Page 102
MSTDEMSTDE with FD smoothing
Euler MSTDE with FD &SD smoothing
Figure 36. Solution depths from MSTDE, MSTDE with frequency / spatialdomain smoothing and Euler (Left), Percentage error in estimated depths(Right).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
103/216
Page 103
Multiple vertical sheets with various separations
Figure 37. Synthetic model data of two vertical sheets, the centres
separated by 100 distance units. Plots of the spatial derivatives, analyticsignal amplitude and various continued tilt products (Top), Synthetic modeldata with MSTD distribution, MSTDE solution density and Euler solutiondensity with MSTDE solution locations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
104/216
Page 104
MSTDE MSTDE at model location
Euler MSTDE AS(VD(TMI))
Figure 38. Solution depths from MSTDE, MSTDE at known model location,MSTDE applied to the AS(VD(TMI)) and Euler (Left), Percentage error inestimated depths (Right).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
105/216
Page 105
Regional / residual separation
The regional field may be removed from the data prior to computation of
the solutions in order to decrease the sensitivity to long wavelength
interference due to proximal sources as well as to portions of sources at
depths significantly deeper than the depth to top of the sources of interest.
Figure 39. Synthetic model data with MSTD distribution, MSTDE solutiondensity and Euler solution density with MSTDE solution locations whenapplied to the total magnetic intensity (Top). The same quantities butapplied to the residual intensity after removal of the regional field (Bottom).The estimated depths are 54 and 51 distance units respectively.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
106/216
Page 106
Contact model
Figure 40. Synthetic model data (Step at a depth of 50 dist. units) with plots
of the spatial derivatives, analytic signal amplitude and various continuedtilt products (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
107/216
Page 107
Figure 41. Synthetic model data (Step at a depth of 50 dist. units) with plotsof the spatial derivatives, analytic signal amplitude and various continuedtilt products (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom). The input data to the algorithm was the AS(VD(TMI)).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
108/216
Page 108
CHAPTER 5: APPLICATION OF MULTI-SCALE DEPTH
ESTIMATION TO OBSERVED DATA AND PERFORMANCE
REVIEW
Field Example 1
Choice of observed data
Due to the method being applicable to two-dimensional, correctly polereduced data, data were chosen over dyke-like features with a near
vertical magnetizing field. Such data was made available over a portion of
the Bushveld Complex in South Africa and may be seen in Figure 43,
which also shows the location of the profile. The data are in a map
projection measured in meters, received in gridded format at a 15 m cell
size and was acquired at a flight height of 50 m above the terrain. All
depths stated herein are relative to the survey platform and not to the
actual ground level. Figure 42 is a schematic of the near surface geology
in the vicinity of the profile.
Figure 42. Schematic representation of the near surface geology of theBushveld in the immediate vicinity of the field data. The thin coversequence (Red) overlies the igneous rocks of the Bushveld Complex (Blue)which have been cross-cut by mafic, magnetically susceptible dykes(Black).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
109/216
Page 109
Figure 43. Pole-reduced total magnetic intensity data over a region of theBushveld Complex in South Africa showing the profile extracted for use inthis example.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
110/216
Page 110
Modelling of observed data
The data were modelled using Encoms ModelVision, Version 11.00,
Release Build 1. Bodies of type tabular were used to represent the likely
sources of the linear magnetic anomalies and their properties optimised
via inversion. The anomaly located at a distance of approximately 6000 m
along the profile was not modelled as it is not sufficiently 2D in nature. The
inducing field intensity used in the modelling was 28370 nT with an
inclination of -63 and a declination of -18, as derived from the available
IGRF information for the region. No remanence was introduced during the
modelling. Table 1 and Figure 44 below illustrate the results obtained fromthis modelling. Figure 45 and Figure 46 illustrate the MSTDE results for
the profile.
Table 1 Positions, widths and susceptibilities of the modelled sources
Distance(m)
Depth(m)
Susceptibility(SI)
Width(m)
Susceptibility x widthproduct
382.6 149.1 0.52 25.2 13.1
1229.6 106.0 0.29 24.9 7.22141.0 72.4 0.17 8.3 1.43517.3 125.5 0.22 11.2 2.54096.8 78.2 0.50 37.8 18.94763.0 49.5 0.47 9.2 4.3
Figure 44. Observed total magnetic intensity data (Black) with regional
(Gray) and modelled intensity (Red, dashed) (Top). Modelled sourcelocations are plotted as x (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
111/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
112/216
Page 112
Figure 46. Observed data with plots of the spatial derivatives, analyticsignal amplitude and various continued tilt products (Top), Observed datawith MSTD distribution, MSTDE solution density and Euler solution densitywith MSTDE solution locations (Bottom). On the bottom subplot, modelled
source locations are plotted as x. The input data to the algorithm was theanalytic signal amplitude of the vertical derivative of the pole-reduced totalmagnetic intensity.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
113/216
Page 113
In Figure 47 it may be seen that the solutions derived from the application
of the MSTDE method to the TMI and to the AS(VD(RTP)) differ to varying
degrees for different anomalies. Some understanding may be gained as to
the geometry of the source from this variation. For step/contact like
features the variation is expected to be larger than for sheet-like sources.
There is generally good agreement between the modelled locations and
those estimated from the MSTDE method.
Figure 47. Histograms of the MSTDE solutions with respect to depth for theapplication to the pole-reduced total magnetic intensity (Top left), The samehistogram but of the application to the AS(VD(RTP)) (Top right), The samehistogram but of the Euler solutions when applied to the pole-reduced totalmagnetic intensity (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
114/216
Page 114
Field Example 2
Choice of observed data
The data used in this example was acquired in 1993 approximately
200 km north-west of Johannesburg, see Figure 48. The region includes
the western limb of the Bushveld Complex in the north-east of the survey
area overlying the floor rocks of the Transvaal Supergroup. The circular
region of high anomalous magnetic intensity due to the Pilanesberg
Complex is located in the centre of the survey area. Three profiles have
been extracted, all in a north-south orientation to be approximately
perpendicular to the mafic dykes that crosscut the region. The data are in
a geographic projection measured in meters; gridded to a 200 m cell size
and while the flight height is not precisely known it was likely in the order
of a few hundred meters (150 m +60 m, Atlas of Magnetic Data - Council
for Geoscience). These data are of significantly poorer quality than those
used in example 1 and thereby serve as a test of the methods ability to
perform in a robust fashion in the presence of an elevated level of noise.
Figure 48. Pole-reduced total magnetic intensity data over a region of theBushveld Complex in South Africa showing the three profiles extracted for
use in this example.
A
B
C
-
7/26/2019 Multi-Scale Tilt Depth Estimation
115/216
Page 115
Results of the application of Multi-Scale Tilt Depth Estimation to
observed data
Figure 49 through Figure 54 illustrate the MSTDE results for profiles A
through C.
Example 2 Profile A
Figure 49. Observed data (Profile A) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Observed data with MSTD distribution, MSTDE solution density and Eulersolution density with MSTDE solution locations (Bottom). The input data tothe algorithm was the total magnetic intensity.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
116/216
Page 116
Figure 50. Observed data (Profile A) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Observed data with MSTD distribution, MSTDE solution density and Eulersolution density with MSTDE solution locations (Bottom). The input data tothe algorithm was the analytic signal amplitude of the vertical derivative ofthe pole-reduced total magnetic intensity.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
117/216
Page 117
Example 2 Profile B
Figure 51. Observed data (Profile B) with plots of the spatial derivatives,
analytic signal amplitude and various continued tilt products (Top),Observed data with MSTD distribution, MSTDE solution density and Eulersolution density with MSTDE solution locations (Bottom). The input data tothe algorithm was the total magnetic intensity.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
118/216
Page 118
Figure 52. Observed data (Profile B) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Observed data with MSTD distribution, MSTDE solution density and Eulersolution density with MSTDE solution locations (Bottom). The input data to
the algorithm was the analytic signal amplitude of the vertical derivative ofthe pole-reduced total magnetic intensity.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
119/216
Page 119
Example 2 Profile C
Figure 53. Observed data (Profile C) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Observed data with MSTD distribution, MSTDE solution density and Eulersolution density with MSTDE solution locations (Bottom). The input data tothe algorithm was the total magnetic intensity.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
120/216
Page 120
Figure 54. Observed data (Profile C) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Observed data with MSTD distribution, MSTDE solution density and Eulersolution density with MSTDE solution locations (Bottom). The input data tothe algorithm was the analytic signal amplitude of the vertical derivative ofthe pole-reduced total magnetic intensity.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
121/216
Page 121
Performance of the Multi-Scale Tilt Depth Estimation
Method
Presented here is a summary of the performance of the MSTDE method
as well as its various implementations during their application to the
synthetic and observed data. The method is sensitive, to differing degrees,
to a number of factors as stated in Chapter 2. The first is the model type.
As seen in the section above, the MSTDE method applied to thin
sources, i.e. with a large depth:width ratio, performs well, however when
applied to thick sources the depths are grossly overestimated or
undefined when the depth to the turning point is sought. In order to return
usable results the method was applied to the AS(VD(TMI)). This approach
returned results slightly too shallow for thin sources but correct for
sufficiently thick sources.
The second factor was the dip of the structure. Above it is shown that the
MSTDE method is quite unstable with respect to varying dip unless the
depth is determined at, or close to, the true location of the upper edge. In
practice this is achieved in two ways: the first is to query the MSTD
distribution at the location of the peak of the analytic signal amplitude as
this is less sensitive to the dip of the structure; the second is to apply the
MSTDE method to the AS(VD(TMI)). This second approach returned
slightly underestimated depths but is relatively stable even to a dip as low
as 10.
The third factor is the sensitivity to the interference due to background
signal. This was tested by applying the MSTDE method to total magnetic
intensity data generated over two proximal sheet type sources. The
distance between these was varied to alter the extent to which the
anomalies interfered. From the results it may be seen that when the
separation ( ) between the sources is within the range 0.5 <
-
7/26/2019 Multi-Scale Tilt Depth Estimation
122/216
Page 122
overestimated. This can be attributed to the apparently lower frequency
due to the interference of the anomalies. If the separation is less than half
of the depth to the sources, the depth is once again well estimated. The
application of the MSTDE method to the AS(VD(TMI)) returned more
stable results in the range 0.5 < 10% of the standard deviation of the
input data. The variable frequency and spatial domain low-pass /
smoothing operations improve the result drastically, allowing the method to
solve for locations however slightly overestimated. An interesting trend
was identified in that the Euler method underestimated the depth whennoise is added. The opposite trend is visible in the low-passed / smoothed
MSTDE solutions. As the added noise amplitude reduces, the solutions
from the MSTDE and Euler methods tend toward the correct depth. This
behavior may be utilized to determine an optimal depth via
experimentation with added noise.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
123/216
Page 123
CHAPTER 6: CONCLUSIONS
The estimation of magnetic source depths from magnetic data is an
extensively studied subject with many, varied approaches adopted. Each
of these approaches has been shown to have particular strengths and
weaknesses with few implemented on a wide scale, commercial basis.
The likes of Euler and Werner deconvolution, shown by Nabighian et al.
(2001) to be unified in three dimensions, are the most prevalent
techniques employed in the modern geophysical industry. This prevalence
does not however preclude the useful application of alternate techniques.
Some comments delivered in this work on previous publications include:
the work presented in Pasteka (2000) on the use of additional polynomial
terms to focus Werner deconvolution solutions on the upper portions of a
source; and the solution strategy adopted by Salem (2003). Comment
delivere d on Pastekas polynomial addition pertains to extending the
concept not only to include low order terms, but to include significantly
higher order terms in order to compensate for high frequency noise
inevitably present within observed data. Salem (2003) makes use ofmultiple levels of continuation, a solution set derived from each. It is found
that solutions are most focused on different sources at different levels of
continuation and the proposal is made that this may be useful for
interpreting large scale datasets where sources of differing natures are
likely to be found. The comment delivered here is that this solution
strategy shows much potential for automation via techniques such as
those of automated clustering presented by Ugalde (2008).
The history of multi-scale computation is somewhat diverse and includes
much work in the signal processing field. Multi-scale computation appears
to have been first applied to geophysical potential fields in 1991 by
McGrath. The focus of this work is however the expression and application
of a newly developed depth estimation technique dubbed Multi-Scale Tilt
Depth Estimation, or here abbreviated as MSTDE. van Buren (2009)
-
7/26/2019 Multi-Scale Tilt Depth Estimation
124/216
Page 124
presents a summary of the method which is elaborated upon here. The
method has been shown to derive suitable depth estimates when applied
to modelled data derived from a range of simple synthetic models.
Sensitivity of the method to model type, dip, interference and noise has
been presented, as well as mitigating strategies to improve and stabilize
the methods performance, which include the application to derivatives of
the magnetic intensity. Code for the computation of the required
parameters and eventual solution sets is included in Appendix A. Results
of the successfully application of the MSTDE method to a field dataset
from the Bushveld Complex in South Africa are presented.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
125/216
Page 125
REFERENCES
Abdelrahman, E. M. and El-Araby, T. M. (2001), A new method for shape
and depth determinations from gravity data: Geophysics, Vol. 66, No. 6,
pp. 1774-1780.
Abdelrahman, E. M. and El-Araby, H. M. (2002), A numerical approach to
depth determination from magnetic data: Kuwait J. Sci. Eng. 29, pp. 121-
134.
Archibald, N. J. and Gow, P. (1999), Multiscale edge analysis of potential
field data: Exploration Geophysics 30, pp. 38-44.
Barbosa, V. C. and Silva, J. B. (1999), Stability analysis and improvement
of structural index estimation in Euler deconvolution: Geophysics, Vol. 64,
No. 1, pp. 48-60.
Blakely, R. J. and Simpson, R. W. (1986), Approximating edges of source
bodies from magnetic or gravity anomalies: Geophysics, Vol. 51, No. 7,
pp. 1494-1498.
Boschetti, F. and Horowitz, F. G. Improved edge detection in potential field
maps and graphical estimation of depth-to-the-top: SEG 70th Annual
Meeting, 2000.Breiner, S. Magnetic Field Derivatives in Geophysical Interpretation: MSc
Thesis, Stanford University, 1962.
Cooper, G. R. J. (2004), Euler Deconvolution applied to Potential Field
Gradients: Exploration Geophysics 35, pp. 165-170.
Cooper, G. R. J. (2004), A semi-automatic procedure for the interpretation
of geophysical data: Exploration Geophysics 35, pp. 180-185.
Cooper, G. R. J. (2008), Euler Deconvolution with Improved Accuracy andMultiple Different Structural indices: Journal of China University of
Geosciences, Vol. 19, No. 1, pp. 72-76.
Cooper, G. R. J. (2004), The stable downward continuation of potential
field data: Exploration Geophysics 35, pp. 260-265.
Dawi, M. G. and Tianyou, L. (2004), Depth Estimation of 2-D Magnetic
Anomalous Sources by Using Euler Deconvolution Method: American
Journal of Applied Sciences 1, pp. 209-214.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
126/216
Page 126
Fairhead, J. D. and Bennett, K. J. Euler: Beyond the "Black Box": SEG
64th Annual Meeting, 1994.
Fedi, M. and Florio, G. Multiridge Analysis and Reduced Euler
Deconvolution : SEG 77th Annual Meeting, 2007.
FitzGerald, D. and Reid, A. (2004), New discrimination techniques for
Euler deconvolution: Computers & Geosciences 30, pp. 461-469.
Florio, G. and Fedi, M. (2006), On the application of Euler deconvolution to
the analytic signal: Geophysics, Vol. 71, No. 6, pp. 87-93.
Florio, G. and Fedi, M. Euler deconvolution of vertical profiles of potential
field data: SEG 76th Annual Meeting, 2006.
Franklin, J. (1990), Analytic Continuation of the Fast Fourier Transform:SIAM Journal of Scientific and Statistical Computing, Vol. 2, No. 1, pp.
112-122.
Hansen, R. O. and Suciu, L. (2002), Multiple-source Euler Deconvolution:
Geophysics, Vol. 67, pp. 525-535.
Hansen, R. O. (2005), 3D multiple-source Werner deconvolution for
magnetic data: Geophysics, Vol 70, No. 5, pp. 45-51.
Holden, D. J. and Archibald, N. J. (2000), Inferring Geological StructuresUsing Wavelet-Based Multiscale Edge Analysis and Forward Models:
Exploration Geophysics 31, pp. 67-71.
Keating, P. and Pilkington, M. A comparison of some interpretation
techniques for magnetic data: SEG 72nd Annual Meeting, 2002.
Keating, P. and Pilkington, M. (2004), Euler deconvolution of the analytic
signal and its application to magnetic interpretation: Geophysical
Prospecting 52, pp. 165-182.Kuttikul, P. Optimization of 3D Euler deconvolution for the interpretation of
potential field data: MSc Thesis, ITC, 1995.
Lahti, I. and Karinen, T. (2010), Tilt derivative multiscale edges of
magnetic data: The Leading Edge, January 2010, pp. 24-29.
Lee, M. and Morris, B. (2010), Effect of signal amplitude on magnetic
depth estimations: The Leading Edge, June 2010, pp. 672-677.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
127/216
Page 127
Lee, M. and Ugalde, H. Spatial Domain Filtering of Specific Frequencies:
Curvature isolation of potential field signals: SEG 78th Annual Meeting,
2008.
Li, X. (2003), On the use of different methods for estimating magnetic
depth: The Leading Edge, November 2003, pp. 1090-1099.
Mallat, S. G. and Zhong, S. Complete Signal Representation with
Multiscale Edges: Robotics Research Technical Report, Computer
Science Division, NYU, 1989.
Mallat, S. G. and Zhong, S. (1992), Characterization of Signals from
Multiscasle Edges: IEEE Transactions on Pattern Analysis and Machine
Intelligence, Vol. 14, No. 7, pp. 710-732.Marson, I. and Klingele, E. E. (1993), Advantages of using the vertical
gradient of gravity for 3-D interpretation: Geophysics, Vol. 58, No. 11, pp.
1588-1595.
Mikhailov, V. and Pajot, G. (2007), Tensor deconvolution: A method to
locate equivalent sources from full tensor gravity data: Geophysics, Vol.
72, No. 5, pp. 161-169.
Miller, H. G. and Singh, V. (1994), Potential field tilt - a new concept forlocation of potential field sources: Journal of Applied Geophysics 32, pp.
213-217.
Mushayandebvu, M. F. and van Driel, P. (2001), Magnetic source
parameters of two-dimensional structures using extended Euler
deconvolution: Geophysics, Vol. 66, No. 3, pp. 814-823.
Mushayandebvu, M. F. and Lesur, V. (2004), Grid Euler deconvolution
with constraints for 2D structures: Geophysics, Vol. 69, No. 2, pp. 489-496.
Nabighian, M. N. and Hansen, R. O. (2001), Unification of Euler and
Werner deconvolution in three dimensions via the generalized Hilbert
transform: Geophysics, Vol. 66, No. 6, pp. 1805-1810.
Nabighian, M. N. (1984), Toward a three-dimensional automatic
interpretation of potential field data via generalized Hilbert transforms:
Fundamental relations: Geophysics, Vol. 49, No. 6, pp. 780-786.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
128/216
Page 128
Pal, D. S. Identification of the Causative Source and Depth from Enhanced
Local Wavenumber Analysis: 6th International Conference & Exposition on
Petroleum Geophysics, 2006.
Pasteka, R. (2006), The Role of the Interference Polynomial in the Euler
Deconvolution Algorithm: Bollettino di Geofisics Teorica ed Applicata, Vol.
47, No. 1-2, pp. 171-180.
Pasteka, R. (2000), 2D Semi-automated Interpretation Methods in
Gravimetry and Magnetometry: ACTA Geologica Universitatis
Comenianae, Nr. 55, pp. 5-50.
Paterson, N. R. and Kwan, K. C. Use of Euler Deconvolution in
Recognizing Magnetic Anomalies of Pipelike Bodies: SEG Extended Abstracts, 1991, pp. 642-645.
Phillips, J. D. Two-step processing for 3D magnetic source locations and
structural indices using extended Euler or analytic signal methods: SEG
72nd Annual Meeting, 2002.
Ravat, D. and Kirkham, K. (2002), A source-depth separation filter: Using
the Euler method on the derivatives of total intensity magnetic anomaly
data: The Leading Edge, April 2002, pp. 360-364.Reid, A. B. and Allsop, J. M. (1990), Magnetic interpretation in three
dimensions using Euler deconvolution: Geophysics, Vol. 55, No 1, pp. 80-
91.
Reid, A. B. (2003), Euler magnetic structural index of a thin-bed fault:
Geophysics, Vol. 68, No. 4, pp. 1255-1256.
Reid, A. B. Semi-Automated Methods of Potential Field Interpretation
Innovations, and Recent and Future Developments: EGM InternationalWorkshop, 2007.
Roest, W. R. and Verhoef, J. (1992), Magnetic interpretation using the 3-D
analytic signal: Geophysics, Vol. 57, No. 1, pp. 116-125.
Salem, A. and Ravat, D. (2003), A combined analytic signal and Euler
method (AN-EUL) for automatic interpretation of magnetic data:
Geophysics, Vol. 68, No. 6, pp. 1952-1961.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
129/216
Page 129
Salem, A. and Ravat, D. (2004), Linearized least-squares method for
interpretation of potential-field data from sources of simple geometry:
Geophysics, Vol. 69, No. 3, pp. 783-788.
Salem, A. and Smith, R. Generalized magnetic tilt-Euler deconvolution:
SEG 72nd Annual Meeting, 2007.
Salem, A. and Williams, S. E. (2008), Interpretation of magnetic data using
tilt-angle derivatives: Geophysics, Vol. 73, No. 1, pp. 1-10.
Salem, A. and Williams, S. E. (2007), Tilt-depth method: A simple depth
estimation method using first-order magnetic derivatives: The Leading
Edge, December 2007, pp. 1502-1505.
Shaw, R. K. and Agarwal, B. N. (2007), Use of Walsh transforms inestimation of depths of idealized sources from total field magnetic
anomalies: Journal of Computers & Geosciences, Vol. 33, No. 7, pp. 966-
975.
Shi-Zhe, X. (2007), A Comparison of Effects Between the Iteration Method
and FFT for Downward Continuation of Potential Fields: Chinese Journal
of Geophysics Vol.50, No.1, pp. 270-274.
Slack, H. A. and Lynch, V. M. (1968), Discussion on T he GeomagneticGradiometer : Geophysics, Vol. 33, No.4 , pp. 680 -683.
Smith, R. S. and Thurston, J. B. (1998), iSPI - the improved source
parameter imaging method: Geophysical Prospecting 46, pp. 141-151.
Stavrev, P. and Gerovska, D. Euler inversion of magnetic anomalies using
analytical continuation: SEG 74th Annual Meeting, 2004.
Stavrev, P. and Reid, A. (2007), Degrees of homogeneity of potential fields
and structural indices of Euler deconvolution: Geophysics, Vol. 72, No. 1,pp. 1-12.
Stavrev, P. and Gerovska, D. (2006), Automatic inversion of magnetic
anomalies from two height levels using finite-difference similarity
transforms: Geophysics, Vol. 71, No. 6, pp. 75-86.
Thompson, D. T. (1982), EULDPH: A new technique for making computer-
assisted depth estimates from magnetic data : Geophysics, Vol. 47, No. 1,
pp. 31-37.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
130/216
Page 130
Thurston, J. B. and Smith, R. S. (2007), Source location using total-field
homogeneity: Introducing the SLUTH method for depth estimation: The
Leading Edge, October 2007, pp. 1272-1277.
Thurston, J. B. and Smith, R. S. (1997), Automatic conversion of magnetic
data to depth, dip, and susceptibility contrast using the SPI (TM) method:
Geophysics, Vol. 62, No. 3, pp. 807-813.
Ugalde, H. and Morris, B. Cluster analysis of Euler deconvolution
solutions: new filtering techniques and actual link to geological structure:
SEG 78th Annual Meeting, 2008.
Valle, M. A. and Keating, P. Magnetic depth estimates using the
continuous wavelet transform: SEG 72nd Annual Meeting, 2002.van Buren, R. Multi-Scale Depth Estimation: 11th SAGA Biennial
Technical Meeting, 2009.
Werner, S. Interpretation of Magnetic Anomalies at Sheet-like Bodies:
Sveriges Geologiska Undersokning, Arsbok 43, No. 6, 1953.
Williams, S. E. and Fairhead, J. D. (2005), Comparison of grid Euler
deconvolution with and without 2D constraints using a realistic 3D
magnetic basement model: Geophysics, Vol. 70, No. 3, pp. 13-21.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
131/216
Page 131
APPENDIX A
Matlab source code for the program developed to compute the Multi-Scale
Tilt Depth estimates follows.% Program to compute Multi-Scale Tilt Depth Estimates % by Reece van Buren, submitted in 2013 to the University% of the Witwatersrand as part of the requirements for the % degree of Master of Science in Geophysics. %% % Environment setup clear format long delete( 'DepthSolutions.csv' ); load( 'MSTDEColourmap.mat' , 'cmap' ) %% % User Input fprintf( '\nEnter the fully qualified name of data file \n' ); fileName = input( ' ... including extension: ' , 's' ); fprintf( '\nPerform AS conversion (0 = No | > 0 = Yes) \n' ); ASC = input( ' ... ' ); ASC=floor(ASC); fprintf( '\nEnter the maximum depth \n' ); zMax = input( ' ... in meters ' ); zMax=abs(fix(zMax)); zMaxNeg=-zMax; levels=floor(zMax/1); % Denominator controls the verticalresolution fprintf( '\nEnter the amplitude of the noise to be added\n' ); fprintf( 'in percent of standard deviation of the data\n' ); ampN = input( ' ... ' ); fprintf( '\nEnter the # of iterations of noise addition andsmoothing\n' ); iter = input( ' ... ' ); if iter1
fprintf( '\nEnter the smoothing increment per iteration (0 =No change)\n' );
deltaSmoothIter = input( ' ... ' ); end fprintf( '\nInitial # of smoothing iterations (0 = No | > 0 = FD &SD) \n' ); smoothIter = input( ' ... ' ); smoothIter = ceil(smoothIter); fprintf( '\nRemove a regional (0 = No | > 0 = Yes) \n' ); regRem = input( ' ... ' ); regRem = ceil(regRem); %% % Read Input Data File Tp = csvread(fileName); %% % Fix Tp length if floor(length(Tp)/2)
-
7/26/2019 Multi-Scale Tilt Depth Estimation
132/216
Page 132
end %% % Setup Variables x = Tp(:,1); Tp = Tp(:,2); n=length(x); n2 = floor(n/2); m2=n2+n; flen=4*n; dx=abs((max(x)-min(x))/(n-1)); fnyq=0.5/dx; df=2*fnyq/(flen-1); f=[-fnyq:df:fnyq]; k=2*pi*f; %% % Reject outlying values cTp=Tp; for i=2:length(Tp)-1
data=Tp(i-1:i+1); med=median(data); sd=std(data); if abs(Tp(i)-med)>sd
cTp(i)=interp1([Tp(i-1) Tp(i+1)],1.5, 'linear' ); end
end Tp=cTp; %% % Interpolate to a fixed data spacing intX=min(x):dx:max(x); int=interp1(x,Tp,intX, 'spline' ); Tp=int';
Tp=Tp-mean(Tp); x=intX'; profLen=max(x)-min(x); origData=Tp; %% % Fix outliers at Tp edges Tp(2)=(Tp(3)-Tp(4))+(Tp(3)+Tp(4))/2; Tp(1)=(Tp(2)-Tp(3))+(Tp(2)+Tp(3))/2; Tp(n-1)=(Tp(n-2)-Tp(n-3))+(Tp(n-2)+Tp(n-3))/2; Tp(n)=(Tp(n-1)-Tp(n-2))+(Tp(n-1)+Tp(n-2))/2; %% % Padding pTp=[zeros(n2+1,1);(-flipud(Tp(2:n))+2*Tp(1)).*((sin([-pi/2:(pi/2)/(n2-1):pi/2])+1)/2)'; Tp; (-flipud(Tp(1:n-1))+2*Tp(n)).*flipud(((sin([-pi/2:(pi/2)/(n2-1):pi/2])+1)/2)');zeros(n2+1,1)]; %% % Remove Regional if regRem>0
fh=exp(-abs(profLen/8)*abs(k)); % UC by 1/8 of profile length pUc=real(ifft(fft(pTp,flen).*fftshift(fh'))); Tp=pTp(m2+1:m2+n)-pUc(m2+1:m2+n); pTp=[zeros(n2+1,1);(-flipud(Tp(2:n))+2*Tp(1)).*((sin([-
pi/2:(pi/2)/(n2-1):pi/2])+1)/2)'; Tp; (-flipud(Tp(1:n-1))+2*Tp(n)).*flipud(((sin([-pi/2:(pi/2)/(n2-1):pi/2])+1)/2)');zeros(n2+1,1)]; end %%
-
7/26/2019 Multi-Scale Tilt Depth Estimation
133/216
Page 133
% Define Taper taper=(sin([-pi/2:(pi/2)/(n-1)*2:pi/2])+1)/2; taper=[taper';ones(length(pTp)-2*length(taper'),1);flipud(taper')]; %% % Set Up Iterations statData=pTp; for q=1:iter
pTp=statData; %%% Add Random Noise for i=1:flen
pTp(i)=(pTp(i)+rand*ampN/100*std(pTp))*taper(i); end %% % Store Copy Of Input Data For Plotting inputData=pTp(m2+1:m2+n); %%
% Perform VDAS Conversion Of Data if ASC>0 % Spatial Smoothing for sm=1:max(smoothIter,1)
pSwop=pTp; for i=2:flen-1
pTp(i)=0.25*pSwop(i-1)+0.5*pSwop(i)+0.25*pSwop(i+1); end
end
% Calculate The HD & VD Of The Data pHd=gradient(pTp,dx); pVd=imag(hilbert(pHd));
% Fix VD Edges Vd=pVd(m2+1:m2+n); Vd(2)=(Vd(3)-Vd(4))+(Vd(3)+Vd(4))/2; Vd(1)=(Vd(2)-Vd(3))+(Vd(2)+Vd(3))/2; Vd(n-1)=(Vd(n-2)-Vd(n-3))+(Vd(n-2)+Vd(n-3))/2; Vd(n)=(Vd(n-1)-Vd(n-2))+(Vd(n-1)+Vd(n-2))/2;
% Repad The VD pVd=[zeros(n2+1,1);(-flipud(Vd(2:n))+2*Vd(1)).*((sin([-
pi/2:(pi/2)/(n2-1):pi/2])+1)/2)'; Vd; (-flipud(Vd(1:n-1))+2*Vd(n)).*flipud(((sin([-pi/2:(pi/2)/(n2-1):pi/2])+1)/2)');zeros(n2+1,1)]; % zero padding
% Calculate The HD & VD Of The VD pHd=gradient(pVd,dx); pVd=imag(hilbert(pHd));
% Calculate The AS Of The VD And Replace The Input Data pTp=sqrt(pVd.^2+pHd.^2);
% Spatial Smoothing pSwop=pTp; for i=2:flen-1
pTp(i)=0.25*pSwop(i-1)+0.5*pSwop(i)+0.25*pSwop(i+1);
end end
-
7/26/2019 Multi-Scale Tilt Depth Estimation
134/216
Page 134
%% % Retaper Output Data pTp=pTp.*taper; Tp=pTp(m2+1:m2+n); %% % Set up 2D Array smpTp=pTp*ones(1,levels); %% % Perform Variable Smoothing if smoothIter > 0 % Perform variable frequency domain smoothing(Preferred)
for Step=1:levels lamda=Step*zMax/levels*2; numOnes=min(floor(1/lamda/df/2),length(f)/2-1); sinVectLen=floor(min(length(f)/2-numOnes,length(f)/16)); sinVect=[0:1/sinVectLen:1]; filt=[ones(numOnes-
1,1);sin(pi/2+sinVect*pi)'/2+0.5;zeros(length(f)/2-numOnes-
sinVectLen,1)]; filtmirror=[filt;flipud(filt)]; smpTp(:,Step)=real(ifft(fft(pTp,flen).*filtmirror));
end if smoothIter>levels % Perform variable spatial domain
smoothing for step = 1:levels
for sm=1:ceil(smoothIter/levels) pSwop=smpTp(:,step); for i=2:flen-1
smpTp(i,step)=0.25*pSwop(i-1)+0.5*pSwop(i)+0.25*pSwop(i+1);
end
end if step0 for sm=1:ceil(smoothIter*step/levels)
pSwop=smpTp(:,step); for i=2:flen-1
smpTp(i,step)=0.25*pSwop(i-1)+0.5*pSwop(i)+0.25*pSwop(i+1);
end end
end end
end end smTp=smpTp(m2+1:m2+n,:); %% % Calculate HD & VD For Each Level for step = 1:levels
pHd(:,step)=gradient(smpTp(:,step),dx); pVd(:,step)=imag(hilbert(pHd(:,step)));
end %% % Calculate 2D Tilt Array
-
7/26/2019 Multi-Scale Tilt Depth Estimation
135/216
Page 135
pTilt=atan(pVd./sqrt(pHd.*pHd)); Vd2D=pVd(m2+1:m2+n,:); Hd2D=pHd(m2+1:m2+n,:); Tilt=pTilt(m2+1:m2+n,:); %% % Calculate Upward-Continuation For Each Level for step = 1:levels
fh=exp(abs(k)*step*zMaxNeg/levels); pUc=real(ifft(fft(pTilt(:,step),flen).*fftshift(fh'))); msDepths(levels+1-step,:)=pUc(m2+1:m2+n);
end %% % Use 10% Smoothed Level To Compute VD & HD For Display Hd=gradient(smpTp(:,ceil(0.1*levels)),dx); Vd=imag(hilbert(Hd)); Hd=Hd(m2+1:m2+n); Vd=Vd(m2+1:m2+n); As=sqrt(Vd.^2+Hd.^2);
%% % Find Zero Contour & Peak Locations zeroc=zeros(1,n); for i=3:n-2
for j=1:levels if msDepths(j,i)>0 & zeroc(i)>-1
zeroc(i)=-levels+j; end
end end j=1; Peak=[0 0]; dzeroc=gradient(zeroc,dx);
for i = 2 : n-2 if zeroc(i)-levels+1 & (dzeroc(i)*dzeroc(i+1)0
smoothIter=smoothIter+deltaSmoothIter; end disp([int2str(ceil(q/iter*100)) '% complete' ]); end %% % Read Back Solutions For Imaging disp( 'Plotting figures' ); SolStatRand = csvread( 'DepthSolutions.csv' ); % The dataDensity Funtion Was Obtained From The Matlab CodeSharing Site SolDens=dataDensity(SolStatRand(:,1),SolStatRand(:,2),(max(x)-min(x))/dx+1,levels,[min(x) max(x) 0 zMaxNeg],dx); %% % Calculate Euler Solutions
-
7/26/2019 Multi-Scale Tilt Depth Estimation
136/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
137/216
Page 137
ylabel( 'nT/dist. unit' ); subplot(5,1,5); axis([min(x) max(x) -Inf Inf]); title( 'Tilt' ); ylabel( 'Radians' ); hold on ; plot(x,msDepths(levels,:), 'Color' ,[0.8 0.8 0.8]); plot(x,msDepths(ceil(levels*0.75),:), 'b' ); %'Color',[0.6 0.60.6]); plot(x,msDepths(ceil(levels*0.5),:), 'g' ); %'Color',[0.3 0.3 0.3]); plot(x,msDepths(1,:), 'r' ); %'Color',[0 0 0]); xlabel( 'Distance' ); box( 'on' );
figure(2); clf(figure(2)); subplot(4,1,1); plot(x,inputData, 'k' ); hold on ; axis([min(x) max(x) -Inf Inf]); title( 'Input Data' ); ylabel( 'nT' ); subplot(4,1,2); imagesc(SolDensStr); hold on ; contourf(flipud(msDepthsStr)); title( 'Multi Scale Tilt Depth Field' ); ylabel( 'Depth' ); grid; axis ij ; axis([min(x) max(x) 0 zMax]);
contour(flipud(msDepthsStr), 'w' ); contour(flipud(msDepthsStr),[0 0], 'w' , 'LineWidth' ,2); subplot(4,1,3); imagesc(sqrt(SolDensStr)); colormap(cmap); hold on ; contour(SolDensStr,[max(max(SolDensStr))/10], '-k' , 'LineWidth' ,1); axis ij ; axis([min(x) max(x) 0 zMax]); plot(intX,zeroc*zMax/levels, 'k' ) title( 'MSTDE Sol. Dens.' ); ylabel( 'Depth' ); grid;
subplot(4,1,4); imagesc(sqrt(EulSolDensStr));title( 'Euler Sol. Dens. & MSTDE Sol.' );xlabel( 'Distance' ); ylabel( 'Depth' ); grid; axis ij ; axis([min(x) max(x) 0 zMax]); hold on ; plot(Peak(:,1)*dx+min(x),-Peak(:,2), 'ok' , 'MarkerSize' ,5); contour(SolDensStr,[max(max(SolDensStr))/10], '-k' , 'LineWidth' ,1);
figure(3);
clf(figure(3));
-
7/26/2019 Multi-Scale Tilt Depth Estimation
138/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
139/216
Page 139
APPENDIX B
Appendix B contains a list of figures (Figure 55 - Figure 89) depicting the
various data used to test the Multi-Scale Tilt Depth Estimation method and
the results thereof. All figures shown within Appendix B make use of the
total magnetic intensity as the input component to the algorithm.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
140/216
Page 140
Vertical sheet at various depths
Figure 55. Synthetic model data (Depth of 5 dist. units) with plots of the
spatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
141/216
Page 141
Figure 56. Synthetic model data (Depth of 10 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
142/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
143/216
Page 143
Figure 58. Synthetic model data (Depth of 30 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
144/216
Page 144
Figure 59. Synthetic model data (Depth of 40 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
145/216
Page 145
Figure 60. Synthetic model data (Depth of 50 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
146/216
Page 146
Figure 61. Synthetic model data (Depth of 60 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
147/216
Page 147
Figure 62. Synthetic model data (Depth of 70 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
148/216
Page 148
Figure 63. Synthetic model data (Depth of 80 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
149/216
Page 149
Sheet with various dips
Figure 64. Synthetic model data (Dip of 10, Depth of 50 dist. units) with
plots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
150/216
Page 150
Figure 65. Synthetic model data (Dip of 20, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
151/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
152/216
Page 152
Figure 67. Synthetic model data (Dip of 40, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
153/216
Page 153
Figure 68. Synthetic model data (Dip of 50, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
154/216
Page 154
Figure 69. Synthetic model data (Dip of 60, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
155/216
Page 155
Figure 70. Synthetic model data (Dip of 70, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
156/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
157/216
Page 157
Figure 72. Synthetic model data (Dip of 90, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
158/216
Page 158
Vertical sheet with various noise amplitudes No smoothing
Figure 73. Synthetic model data (Noise range of 10% of the standard
deviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Nosmoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
159/216
Page 159
Figure 74. Synthetic model data (Noise range of 25% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). NoSmoothing has been applied to the data. No smoothing has been applied tothe data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
160/216
Page 160
Figure 75. Synthetic model data (Noise range of 50% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Nosmoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
161/216
Page 161
Figure 76. Synthetic model data (Noise range of 100% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Nosmoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
162/216
Page 162
Vertical sheet with various noise amplitudes Frequency domain
smoothing
Figure 77. Synthetic model data (Noise range of 10% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain smoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
163/216
Page 163
Figure 78. Synthetic model data (Noise range of 25% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain smoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
164/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
165/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
166/216
Page 166
Vertical sheet with various noise amplitudes Frequency domain
and 100 iterations of spatial domain smoothing
Figure 81. Synthetic model data (Noise range of 10% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain and up to 100 iterations of spatial domain smoothing have beenapplied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
167/216
Page 167
Figure 82. Synthetic model data (Noise range of 25% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain and up to 100 iterations of spatial domain smoothing have beenapplied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
168/216
Page 168
Figure 83. Synthetic model data (Noise range of 50% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain and up to 100 iterations of spatial domain smoothing have beenapplied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
169/216
Page 169
Figure 84. Synthetic model data (Noise range of 100% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain and up to 100 iterations of spatial domain smoothing have beenapplied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
170/216
Page 170
Multiple vertical sheets with various separations
Figure 85. Synthetic model data (Separation of 20 dist. units, Depth of 50
dist. units) of two vertical sheets, the centres separated by 100 distanceunits. Plots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
171/216
Page 171
Figure 86. Synthetic model data (Separation of 50 dist. units, Depth of 50dist. units) of two vertical sheets, the centres separated by 100 distanceunits. Plots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
172/216
Page 172
Figure 87. Synthetic model data (Separation of 100 dist. units, Depth of 50dist. units) of two vertical sheets, the centres separated by 100 distanceunits. Plots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
173/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
174/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
175/216
Page 175
APPENDIX C
Appendix C contains a list of figures (Figure 90 - Figure 124) depicting the
various data used to test the Multi-Scale Tilt Depth Estimation method and
the results thereof. All figures shown within Appendix C make use of the
analytic signal amplitude of the vertical derivative of the total magnetic
intensity as the input component to the algorithm.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
176/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
177/216
Page 177
Figure 91. Synthetic model data (Depth of 10 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
178/216
Page 178
Figure 92. Synthetic model data (Depth of 20 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
179/216
Page 179
Figure 93. Synthetic model data (Depth of 30 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
180/216
Page 180
Figure 94. Synthetic model data (Depth of 40 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
181/216
Page 181
Figure 95. Synthetic model data (Depth of 50 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
182/216
Page 182
Figure 96. Synthetic model data (Depth of 60 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
183/216
Page 183
Figure 97. Synthetic model data (Depth of 70 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
184/216
Page 184
Figure 98. Synthetic model data (Depth of 80 dist. units) with plots of thespatial derivatives, analytic signal amplitude and various continued tiltproducts (Top), Synthetic model data with MSTD distribution, MSTDEsolution density and Euler solution density with MSTDE solution locations(Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
185/216
Page 185
Sheet with various dips
Figure 99. Synthetic model data (Dip of 10, Depth of 50 dist. units) with
plots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
186/216
Page 186
Figure 100. Synthetic model data (Dip of 20, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
187/216
Page 187
Figure 101. Synthetic model data (Dip of 30, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
188/216
Page 188
Figure 102. Synthetic model data (Dip of 40, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
189/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
190/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
191/216
Page 191
Figure 105. Synthetic model data (Dip of 70, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
192/216
Page 192
Figure 106. Synthetic model data (Dip of 80, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
193/216
Page 193
Figure 107. Synthetic model data (Dip of 90, Depth of 50 dist. units) withplots of the spatial derivatives, analytic signal amplitude and variouscontinued tilt products (Top), Synthetic model data with MSTD distribution,MSTDE solution density and Euler solution density with MSTDE solutionlocations (Bottom).
-
7/26/2019 Multi-Scale Tilt Depth Estimation
194/216
Page 194
Vertical sheet with various noise amplitudes No smoothing
Figure 108. Synthetic model data (Noise range of 10% of the standard
deviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Nosmoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
195/216
Page 195
Figure 109. Synthetic model data (Noise range of 25% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). NoSmoothing has been applied to the data.. No smoothing has been appliedto the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
196/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
197/216
Page 197
Figure 111. Synthetic model data (Noise range of 100% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Nosmoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
198/216
Page 198
Vertical sheet with various noise amplitudes Frequency domain
smoothing
Figure 112. Synthetic model data (Noise range of 10% of the standard
deviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain smoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
199/216
Page 199
Figure 113. Synthetic model data (Noise range of 25% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain smoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
200/216
Page 200
Figure 114. Synthetic model data (Noise range of 50% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain smoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
201/216
Page 201
Figure 115. Synthetic model data (Noise range of 100% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain smoothing has been applied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
202/216
-
7/26/2019 Multi-Scale Tilt Depth Estimation
203/216
Page 203
Figure 117. Synthetic model data (Noise range of 25% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain and up to 100 iterations of spatial domain smoothing have beenapplied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
204/216
Page 204
Figure 118. Synthetic model data (Noise range of 50% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain and up to 100 iterations of spatial domain smoothing have beenapplied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
205/216
Page 205
Figure 119. Synthetic model data (Noise range of 100% of the standarddeviation, Depth of 50 dist. units) with plots of the spatial derivatives,analytic signal amplitude and various continued tilt products (Top),Synthetic model data with MSTD distribution, MSTDE solution density andEuler solution density with MSTDE solution locations (Bottom). Frequencydomain and up to 100 iterations of spatial domain smoothing have beenapplied to the data.
-
7/26/2019 Multi-Scale Tilt Depth Estimation
206/216
P